Contents lists available at ScienceDirect

Landscape and Urban Planning

journal homepage: www.elsevier.com/locate/landurbplan

Research Paper

Participatory soundscape sensing

Chunming Lia,⁎, Yin Liub, Muki Haklayc

a Key Lab of Urban Environment and Health, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, Chinab Xiamen Research Academy of Environmental Sciences, Xiamen 361004, Chinac Extreme Citizen Science Group, University College London, Gower Street, London WCIE 6BT, United Kingdom

A R T I C L E I N F O

Keywords:SoundscapeParticipatory sensingEnvironmental noiseUrbanCitizen science

A B S T R A C T

Soundscape research offers new ways to explore the acoustic environment and potentially address challenges. Acomprehensive understanding of soundscape characteristics and quality requires efficient data collection andanalysis methods. This paper describes Participatory Soundscape Sensing (PSS), a worldwide soundscape in-vestigation and evaluation project. We describe the calibration method for sound pressure levels (SPL) measuredby mobile phone, analyze the PSS’s data temporal-spatial distribution characteristics, and discuss the impact ofthe participants’ age and gender on the data quality. Furthermore, we analyze the sound comfort level re-lationships with each class of land use, sound sources, subjective evaluation, sound level, sound harmoniousness,gender, and age using over a year of shared data. The results suggest that PSS has distinct advantages in en-hancing the amount and coverage of soundscape data. The PSS data distribution is closely related to the tem-poral pattern of the human work-rest schedule, population density, and the level of cyber-infrastructure. Adults(19–40 years old) are higher-quality data providers, and women exhibit better performance with respect to dataintegrity than men. Increasing the proportion of natural source sounds and reducing the proportion of human-made sources of sound is expected to enhance the sound comfort level. A higher proportion of sound harmo-niousness leads to higher sound comfort, and the higher proportion of subjective evaluation sound level does notlead to decreased sound comfort. We suggest that the crowdsourcing data with participatory sensing will providea new perspective in soundscape investigation, evaluation, and planning.

1. Introduction

Soundscape can be defined as the acoustic environment perceived,experienced, and/or understood by a person or people in a given con-text (ISO 12913-1, 2014), which places emphasis on the perception,evaluation, and experience of the listeners. The urban soundscape ap-proach considers the acoustic environment as a “resource” (Brown,2012) with the goal of improving urban sound quality via design andplanning. The main topics of the urban soundscape include soundsource identification (Jeon & Hong, 2015), spatial-temporal variation(Hong & Jeon, 2017; Liu, Kang, Luo, Behm, & Coppack, 2013), in-dicators selection (Aletta, Kang, & Axelsson, 2016), sound evaluation(Yang & Kang, 2005; Zhang, Zhang, Liu, & Kang, 2016), and soundscapedesign (Chung, To, & Schulte-Fortkamp, 2016). Soundscape researchmethods, including pen and paper questionnaires, interviews, soundwalks, and replaying of sound records in the lab, have been used tocollect data, such as sound sources, sound pressure levels, location in-formation, individual feelings, and demographic factors, among others(He & Pang, 2016; Kang, 2014; Liu, Kang, Behm, & Luo, 2014), and

most of these factors have significant costs and time investment. Labtests mean that volunteers cannot feel the real soundscape directly and,moreover, a long test can easily tire the participants. As a result, currentresearch projects are primarily conducted at a small scale, such as in apark or green space, which leads to results that are difficult to apply ona large scale. Because soundscape design includes multi-party partici-pation and discussion, reasonable soundscape design requires addi-tional participants (He & Pang, 2016).

Participatory sensing (PS) is the process through which individualsand communities use the capabilities of mobile devices and cloud ser-vices to collect, analyze, and contribute sensory information (Burkeet al., 2006; Estrin et al., 2010). Using the concept of PS, sound-re-cording and noise-monitoring mobile applications and online websurvey software have been reported. Noteworthy is that some mobilephones’ accuracy for measuring noise pollution has been tested(Aumond et al., 2017), but few of them may be appropriate for noisemeasurement (Kardous & Shaw, 2014). The soundscape quality-relatedinformation, including such factors as sound pressure level (SPL), soundfrequency, land use, or subjective evaluation, cannot be completely

https://doi.org/10.1016/j.landurbplan.2018.02.002Received 7 November 2017; Received in revised form 9 February 2018; Accepted 10 February 2018

⁎ Corresponding author.E-mail address: cmli@iue.ac.cn (C. Li).

Landscape and Urban Planning 173 (2018) 64–69

0169-2046/ © 2018 Elsevier B.V. All rights reserved.

T

recorded (Becker et al., 2013; Cordeiro, Barbosa, & Afonso, 2013; Craig,Moore, & Knox, 2017; Drosatos, Efraimidis, Athanasiadis, Stevens, &Hondt, 2014; Hedfors, 2013; Yelmi, Kuscu, & Yantac, 2016). Ad-ditionally, the quality and characteristics of these crowdsourced datalack detailed descriptions or discussion.

In this paper, we propose Participatory Soundscape Sensing (PSS),which is an ongoing worldwide soundscape investigation and evalua-tion project that engages the public in participatory sensing. We de-scribe the PSS tools and the calibration method of SPL as measured bymobile phones. We analyze the temporal-spatial distribution char-acteristics of the PSS data; discuss the impact of the participants’ ageand gender on the quality of data, including length of measurementtime and soundscape records integrity; and analyze the sound comfortlevel relationships with each class of land use, sound sources, subjectiveevaluation sound level, sound harmoniousness, gender, and age.

2. Methodology

2.1. PSS tools development

The PSS tools include SPL Meter and PSS Server. SPL Meter (whichcan be downloaded at http://www.citi-sense.cn/download) is asoundscape data investigation and analysis software package that canbe installed on both Android and iOS operating systems. PSS Serverruns on a cloud server and can analyze and visualize soundscape dataonline from around the world (http://pss.citi-sense.cn).

Fig. 1 shows the logical architecture of SPL Meter contains fourmain components, including SPL calculation, location and sound sourceidentification, demographic information and time collection, and re-sults storage and sharing.

2.1.1. SPL calculationA continuous signal can be adequately sampled only if it contains

frequency components greater than one-half of the sampling rate(Smith, 1999). The average human ear senses tones resulting fromsound oscillation at frequencies between 20 and 20,000 Hertz (Hz), andthe most sensitive frequencies span the range of 2000–5000 Hz. SPLMeter receives 16-bit PCM (pulse-code modulation is a digital re-presentation of an analogue signal) at a speed of 44,100 Hz from itsmicrophone. SPL Meter extracts the amplitude and frequency from thesampled signal using the Fast Fourier Transformation (FFT). For the

purpose of this application, the calculation method of FFT comes fromthe ddf.minim.analysis package and the block size was set as 2048 inFFT. The human ear does not respond to these frequencies equally welland is less sensitive to extreme high and low frequencies; therefore, anA-weighted SPL, which is modified by the A-weighting filter, is com-monly used in noise dose measurement at work. The A-weightedequivalent continuous sound level (LAeq), maximum sound level (mLpa)and its corresponding frequency (mF), the sound level exceeded for 10%of the time of the measurement duration (L10), the sound level exceededfor 50% of the time of the measurement duration (L50), and the soundlevel exceeded for 90% of the time of the measurement duration (L90)can be calculated using A-weighted SPL. The calculation results areshown on the main screen of the SPL Meter by numeric representationor as a graph.

2.1.2. Location and sound source identificationDifferences in land use and sound sources can affect the perception

of the soundscape (Kang, 2007). The information for land use andsound sources can be identified by the participants using a list in theevaluation interface of the SPL Meter app. The latest list of land use andsound sources is supplied when SPL Meter connects to PSS Server eachtime it starts. Each item of the land use and sound sources has a uniquecode. The lists are updated if new items (sound source or land use in-formation) are added to the lists in PSS server. The location coordinatesare collected using the mobile phone’s high-accuracy location service(GPS, WLAN, or mobile networks).

2.1.3. Soundscape evaluationThe subjective evaluation of sound levels, sound comfort levels, and

sound harmoniousness levels, which are widely used in soundscapeevaluation (Aspuru, García, Herranz, & Santander, 2016; Kang, 2007),can also be applied in SPL Meter, where each is divided into five linearscales that were standardized in noise surveys (Fields et al., 2001). Thelevel of harmonization between aural and visual perception has beendefined as sound harmoniousness level in this study. Information re-lated to the gender and age of the participants can also be collected ifthe user is willing to supply them. The local time, time zone, and UTCare obtained when SPL Meter is used to measure and evaluate thesoundscape.

The state of the earphone is necessary to judge whether the internalor external microphone is used. Other hardware and software variations

Fig. 1. Logical architecture for SPL Meter.

C. Li et al. Landscape and Urban Planning 173 (2018) 64–69

65

might exist if an external microphone of unknown properties is used,but we can expect that most mobile phones’ internal microphone ty-pically has a sensitivity of -50 dB. The notification of the PSS server canbe shown on the top of the main interface, which is useful for PSSproject maintenance. Measurements can be stored in the mobile phonesor they can be shared with the PSS server.

2.1.4. Participatory results visualizationReal-time measurements are submitted by the participants and

analyzed on PSS server, and the subsequent analytical results are illu-strated on the website using maps, pie charts, and histograms.Information on the interface includes the number of total participantsand records; the proportion of place types, sound sources, age, andgender; and evaluation of sound level, sound comfort, and sound har-moniousness. The media of SPL, maxSPL, equivalent SPL, and fre-quency are also presented on the web page, as illustrated in Fig. 2.

2.2. SPL data calibration

The sensitivity of the mobile phone microphone is much lower thanthat of a purpose-built sound meter (e.g., the sensitivity of HS5633T is-31.7 dB). Microphones from different mobile phone companies havedifferent sensitivities and should be calibrated before measuring theSPL. Kardous and Shaw (2014) used pink noise with a 20–20,000 Hzfrequency range, at levels from 65 dB to 95 dB, and Aumond et al.(2017) used white noise from 35 dBA to 100 dBA to calibrate theirmobile phones. In this study, firstly, four different model types of mo-bile phones equipped with SPL Meter and a sound pressure meter (SPM)(HS5633T/Heng Sheng Electronics) that meet the National VerificationRegulation of Sound Level Meters (JJG188-2002) were put together inthe same sound field. The distance between the phones’ microphoneand the speaker was 1m. Secondly, we generated different frequencynoise with 20–20,000 Hz noise to test our phones and SPM at the sametime, and calculated the correlation parameters with SPM at levels from35 dBA to 90 dBA using the linear regression method. Finally, thesecalibrated mobile phones were used outdoors to measure the equivalentSPL three times, with each measurement lasting for 20min. Ad-ditionally, a 94 dBA consistent sound source device (HS6020/HengSheng Electronics) was used before and after each measurement to

control the error of SPM (not exceeding 0.5 dBA).

2.3. Data quality analysis

After more than a year of operation (from March 1st, 2016 to August31st, 2017), we obtained the PSS data temporal variation, spatial dis-tribution and accuracy of GPS, and analyzed the participants’ age andgender impacts on the data quality, including the ratio of sharedmeasurements, length of measurement time, and integrity of measure-ment records. The records integrity describes the proportion of eachsoundscape related indictor recorded: for example, if there are 50 GPSrecords in 100 measurement activities, the integrity of GPS indicators is50%. In addition, we analyzed the sound comfort level relationshipswith each class of land use, sound sources, subjective evaluation soundlevel, sound harmoniousness, gender, and age.

3. Results and discussion

3.1. SPL data validation

During the study period, we received observations from 470 modeltypes belonging to 45 mobile phone manufacturers. Certain models thatwe have were calibrated, while others can be calibrated in a similarmanner. Fig. 3 shows that these mobile phones have good correlationwith SPM. Table 1 shows the average error between each of the mobilephones and SPM is 0.3 dBA (HTC Desire), 0.8 dBA (HTC Wildfire),1.2 dBA (HTC Incredible), and 0.7 dBA (SAMSUNG I9000), meaningthat the calibrated mobile phones are suitable for measuring SPL.

3.2. Data temporal-spatial distribution

The number of participants has continuously increased since therelease of SPL Meter on the app market (e.g., Google Play, iTunes,Baidu, QQ, anzhi, etc.) in March 2016. Over 11,326 downloads wererecorded at the end of August 2017, and approximately 5601 partici-pants shared 25,471 measurement records. Wi-Fi is the main channelfor data sharing (Android: 60.78%, IOS: 64.22%). Fig. 4 shows thatmeasurements were mainly concentrated from 9:00am to 11:00pm,which is closely related to the temporal pattern of the human work-rest

Fig. 2. PSS online analysis and visualization website.

C. Li et al. Landscape and Urban Planning 173 (2018) 64–69

66

schedule. The number of women was less than the number of men(women: 9.8%, men: 90.2%), which may explain why the daily varia-tion of women is uneven.

The measurement sites gradually spread around the world at theend of August 2017, as indicated in Fig. 5. Numerous populations,ubiquitous networks, and plentiful numbers of mobile applicationmarkets make the measurement sites in China and USA much morenumerous than in other locations.

3.3. Data quality impacted by gender and age

A complete measurement record includes information on LAeq, mLpa,mF, L10, L50, L90, land use, GPS, gender, age, sound sources, and sub-jective sound evaluation level (level, comfort, and harmoniousness).The first six physical indicators described the sound and are not im-pacted by the participants’ demographic biases. The subjectivesoundscape evaluation, sound sources, and class of land use identifi-cation, which require knowledge other than gender and age, are unevenin the differences among participants’ demographic biases. Fig. 6 showsthe record integrity for participants under 12 years old was much lowerthan that of other age groups. Women show better performance in data

integrity (completing the recording) than men. The accuracy of GPS iseasily affected by the surroundings, but most distances (81.5%) are lessthan 50m.

The longer the measurement time, the more meaningful the resultsare. Fig. 7 shows the length of each use of SPL Meter time was mainly(90.3%) concentrated in the range of 10–101 s and half of the mea-surement activities (50.7%) were initiated by participants 19–40 yearsof age. The ratio of participants whose ages are under 12 years olddecreased most rapidly with increased measurement time as shown inFig. 7, which suggests that these participants have more difficulty insupplying richer records than the other age groups. The percentage ofmen was significantly higher than women in the different measurementtime (The p-value is 0.006 in t-Test at p < 0.01 level).

3.4. Sound comfort evaluation

When the sound comfort level is shifted from very uncomfortable tovery comfortable, Fig. 8 shows the proportion of natural sources con-tinuously increases (from 15.23% to 41.02%) and the proportion ofhuman-made sources continuously decreases (from 68.42% to 36.21%),but the proportion of music (which is one of the human-made sourcesounds) increases. The proportion of human activity sources increasesfrom a very uncomfortable level to a comfortable level but decreases atthe very comfortable level. Water sounds are the most likely to makepeople feel more sound comfortable as compared to other naturalsource sounds. In addition, machine sound is the most unwelcomesound.

Most of the measurement activities were conducted in a residentialarea (R). The categories of business area (B), industrial area (M), androad, street, and transportation area (S) have lower proportions at thehighest sound comfort level.

Based on the results, we find that increasing the proportion ofnatural source sounds and more reasonable land use configurations thatreduce the proportion of human-made source sounds can be expected toenhance the sound comfort level. However, increasing human activitiessource sound does not decrease sound comfort.

When the sound comfort level is shifted from very uncomfortable tovery comfortable, Fig. 9 shows the sound harmoniousness level is alsoenhanced, whereas the subjective evaluation sound level does not de-crease, which means that the sound harmoniousness levels are morevaluable than the subjective evaluation of sound level.

In addition, we find that, when the sound comfort level is shiftedfrom very comfortable to very uncomfortable, the ratio of participantsthat are women and older than 60 years continuously increases. Thewomen’s ratio increased by a factor of five (from 4.22% to 22.54%),and the ratio for the age group older than 60 years increased by 7%(from 2.21% to 9.21%). The results show that elderly people andwomen may be more easily negatively affected by environmental noise.

4. Conclusions

PSS assigns the task of standardized data collection and calculationto citizens around the world with the aid of SPL Meter and mobilenetworks. Citizens can be involved at any time and any location withtheir smart devices, and a long-term research network can be easily andquickly formed with more participants, which is highly useful for im-proving data collection efficiency and accumulating large data sets forsoundscape research, design, and planning.

The PSS data temporal-spatial distribution is closely related to thetemporal pattern of the human work-rest schedule, population density,and level of cyber-infrastructure. The data quality primarily depends onthe knowledge of the individuals or communities and the capabilities oftheir devices, which is different from that of data from questionnairesguided by interviewers in situ. Rich and specific classification of soundsources and land use is expected to supply more valuable data, but itmight decrease the user experience and lead to complicated operation

Fig. 3. Different mobile phones compared with SPM.

Table 1The LAeq values of SPM and mobile phones in the same outdoor environment (dBA).

ID SPM Error ofSPM(before,after)

HTCDesire(error)

HTCWildfire(error)

HTCIncredible(error)

SAMSUNGI9000(error)

1 49.2 0.4 (94.3,93.9)

49.5 (0.3) 49.8 (0.6) 50.4 (1.2) 49.9 (0.7)

2 49.0 0.1 (94.2,94.3)

49.2 (0.2) 49.9 (0.9) 50.2 (1.2) 49.5 (0.5)

3 49.1 0.4 (94.2,94.6)

49.4 (0.3) 50.0 (0.9) 50.4 (1.3) 50.1 (1.0)

Fig. 4. Daily variation of measured activities.

C. Li et al. Landscape and Urban Planning 173 (2018) 64–69

67

or even abandonment of the tools. As a result, the question of how tohelp citizens from different cultures and knowledge levels to under-stand the terminology and to simplify and standardize the operation ofsoftware and devices will be a great challenge in the future.

Because the sound comfort level has a close relationship with de-mographic biases and land use, sound pressure level control is an im-portant method used to improve the sound comfort level, whereas othermethods, including enhancing the ratio of natural source sounds (water,insects, etc.), more reasonable land use configurations to reduce the

Fig. 5. Map of measured sites and percentages in each country.

Fig. 6. Gender and age impacts on record integrity.

Fig. 7. Gender and age impacts on measured time.

Fig. 8. Percentage of sound sources and land uses at different sound comfort levels.

Fig. 9. Percentage of subjective evaluation sound level and sound harmoniousness atdifferent sound comfort levels.

C. Li et al. Landscape and Urban Planning 173 (2018) 64–69

68

ratio of human-made source sounds, and enhancing the sound harmo-niousness level, are expected to be helpful in improving the soundcomfort level.

Acknowledgments

This work was supported by the National Natural ScienceFoundation of China [Grant No. 41301577], the Youth InnovationPromotion Association, Chinese Academy of Sciences [Grant No.2017351], the Young Talents Program, Chinese Academy of Sciences[Grant No. IUEQN201303], and the Fujian Science and TechnologyProjects [Grant No. 2016Y0081].

References

Aletta, F., Kang, J., & Axelsson, Ö. (2016). Soundscape descriptors and a conceptualframework for developing predictive soundscape models. Landscape and UrbanPlanning, 149, 65–74. http://dx.doi.org/10.1016/j.landurbplan.2016.02.001.

Aspuru, I., García, I., Herranz, K., & Santander, A. (2016). CITI-SENSE: methods and toolsfor empowering citizens to observe acoustic comfort in outdoor public spaces. NoiseMapping, 3, 37–48. http://dx.doi.org/10.1515/noise-2016-0003.

Aumond, P., Lavandier, C., Ribeiro, C., Boix, E. G., Kambona, K., D’Hondt, E., & Delaitre,P. (2017). A study of the accuracy of mobile technology for measuring urban noisepollution in large scale participatory sensing campaigns. Applied Acoustics, 117(PartB), 219–226. http://dx.doi.org/10.1016/j.apacoust.2016.07.011.

Becker, M., Caminiti, S., Fiorella, D., Francis, L., Gravino, P., Haklay, M., et al. (2013).Awareness and learning in participatory noise sensing. PLoS One, 8(12), e81638.http://dx.doi.org/10.1371/journal.pone.0081638.

Brown, A. L. (2012). A review of progress in soundscapes and an approach to soundscapeplanning. International Journal of Acoustics and Vibration, 17(2), 73–81. http://hdl.handle.net/10072/50262.

Burke, J.A., Estrin, D., Hansen, M., Parker, A., Ramanathan, N., Reddy, S., & Srivastava,M.B. (2006). Participatory Sensing. http://escholarship.org/uc/item/19h777qd/(Accessed 14 March 2017).

Chung, A., To, W. M., & Schulte-Fortkamp, B. (2016). Next generation soundscape designusing virtual reality technologies. The Journal of the Acoustical Society of America,140(4), http://dx.doi.org/10.1121/1.4969442 3041-3041.

Cordeiro, J., Barbosa, Á., & Afonso, B. (2013). Soundscape-Sensing in Social Networks.AIA-DAGA 2013 Proceedings of Conference on AcousticsMerano: EAA Euroregiohttp://www.abarbosa.org/docs/Barbosa_AIADAGA-2013.pdf.

Craig, A., Moore, D., & Knox, D. (2017). Experience sampling: Assessing urban sounds-capes using in-situ participatory methods. Applied Acoustics, 117(Part B), 227–235.http://dx.doi.org/10.1016/j.apacoust.2016.05.026.

Drosatos, G., Efraimidis, P. S., Athanasiadis, I. N., Stevens, M., & Hondt, E. D. (2014).

Privacy-preserving computation of participatory noise maps in the cloud. Journal ofSystems and Software, 92, 170–183. http://dx.doi.org/10.1016/j.jss.2014.01.035.

Estrin, D., Chandy, K. M., Young, R. M., Smarr, L., Odlyzko, A., Clark, D., et al. (2010).Participatory sensing: applications and architecture. IEEE Internet Computing, 14,12–42. http://dx.doi.org/10.1109/MIC.2010.12.

Fields, J. M., DeJong, R. G., Gjestland, T., Flindell, I. H., Job, R. F. S., Kurra, S., et al.(2001). Standardized general-purpose noise reaction questions for community noisesurveys: research and a recommendation. Journal of Sound and Vibration, 242(4),641–679. http://dx.doi.org/10.1006/jsvi.2000.3384.

He, M., & Pang, H. (2016). A review of soundscape research history and progress.Landscape Architecture, 88–97. https://doi.org/10.14085/j.fjyl.2016.05.0088.10.

Hedfors, P.. Mobile app for the characterization of soundscapes. (2013). http://www.slu.se/en/departments/urban-rural-development/research/landscape-architecture/projects/soundscapes/ (Accessed 14 March 2017).

Hong, J. Y., & Jeon, J. Y. (2017). Exploring spatial relationships among soundscapevariables in urban areas: A spatial statistical modelling approach. Landscape andUrban Planning, 157, 352–364. http://dx.doi.org/10.1016/j.landurbplan.2016.08.006.

International Organization for Standardization. ISO 12913-1:2014 Acoustics – Soundscape –Part 1: Definition and Conceptual Framework. (2014). https://www.iso.org/obp/ui/#iso:std:iso:12913:-1:ed-1:v1:en/ (Accessed 14 March 2017).

Jeon, J. Y., & Hong, J. Y. (2015). Classification of urban park soundscapes through per-ceptions of the acoustical environments. Landscape and Urban Planning, 141,100–111. http://dx.doi.org/10.1016/j.landurbplan.2015.05.005.

Kang, J. (2007). Urban sound environment (1st ed.). New York: Taylor & Francis.Kang, J. (2014). Soundscape: Current progress and future development. New

Architecture, 4–7.Kardous, C. A., & Shaw, P. B. (2014). Evaluation of smartphone sound measurement

applications. The Journal of the Acoustical Society of America, 135, EL186. http://dx.doi.org/10.1121/1.4865269.

Liu, J., Kang, J., Behm, H., & Luo, T. (2014). Effects of landscape on soundscape per-ception: Soundwalks in city parks. Landscape and Urban Planning, 123, 30–40. http://dx.doi.org/10.1016/j.landurbplan.2013.12.003.

Liu, J., Kang, J., Luo, T., Behm, H., & Coppack, T. (2013). Spatiotemporal variability ofsoundscapes in a multiple functional urban area. Landscape and Urban Planning, 115,1–9. http://dx.doi.org/10.1016/j.landurbplan.2013.03.008.

Smith, S. W. (1999). The scientist and engineer’s guide to digital signal processing (2nd ed.).California: California Technical Pub.

Yang, W., & Kang, J. (2005). Acoustic comfort evaluation in urban open public spaces.Applied Acoustics, 66(2), 211–229. http://dx.doi.org/10.1016/j.apacoust.2004.07.01.

Yelmi, P., Kuscu, H., & Yantac, A.E. (2016). Towards a sustainable crowdsourced soundheritage archive by public participation: the soundsslike project. In The 9th NordicConference on Human-Computer Interaction, Sweden. 1–9. doi: http://dx.doi.org/10.1145/2971485.2971492.

Zhang, D., Zhang, M., Liu, D., & Kang, J. (2016). Soundscape evaluation in Han ChineseBuddhist temples. Applied Acoustics, 111, 188–197. http://dx.doi.org/10.1016/j.apacoust.2016.04.020.

C. Li et al. Landscape and Urban Planning 173 (2018) 64–69

69